Spelling out the future of food, with IIoT and AI. By Mike Edgett
From timings to temperatures and miles travelled, food production creates masses of data. Most businesses waste this information, but as the Industrial Internet of Things (IIoT) makes its presence felt, those who do discard this data, face extinction.
The numbers behind IIoT are staggering. The total installed base of IIoT connected devices is projected to reach more than 75 billion worldwide by 2025. This is a fivefold increase in ten years, creating some 25 billion connections.
Whilst much of this early growth will be consumer devices, many analysts predict that after 2025, the IIoT will begin to overshadow the consumer variation. The industrial segment – which encompasses both IoT technology deployed within enterprises and vertical-specific applications – will make up more than half of the connections, reaching just under 14 billion.
This is data on a truly awesome scale. But within food and beverage particularly, data must be consumed quickly to be relevant and useful and the truth remains that many manufacturers often struggle to process and apply data insights quickly enough. The IIoT data overload will only make things worse, unless the industry begins to adapt now. The question is of course, how?
Taming IIoT data
Data is the driving force behind commercial success in the 21st century. From demand forecasts to production schedules and traceability or new customer buying trends, the principle attraction of the IIoT is to expose data and insight that was previously hidden.
Hundreds and thousands of sensors generate massive volumes in a dynamic, ceaseless flow of data points. Organisations are often ill equipped to aggregate, prioritise, and draw conclusions. Even simple storage of the volumes of data can be a challenge, forcing companies to turn to cloud computing with elastic flexibility.
This is partly because of a lack of expertise. IT professionals with data science skills are in very short supply – especially for businesses located outside of technology hubs. Companies have often been forced to turn to third parties which can become costly.
Consequently, artificial intelligence (AI) has quickly become a key tool in addressing this overload. AI-driven analytics can forecast likely outcomes and prescribe responses much quicker than most data scientists. However, there remains an issue of time – typically an industrial AI project requires a substantial lead time and faces the risks of still having to interact with human skills such as report writing.
Making AI human
The key is to ensure that the technology that addresses this IIoT data is as user-friendly and intuitive as possible. There is a new breed of AI tools that puts powerful predictive analytics in the hands of front-line users, helping them address day-to-day needs with greater insight. It’s no longer necessary to turn to code-writing developers to create use-case-specific applications. Users should be able to just delve into the data they care about and analyse it how they determine.
Consequently, modern AI toolsets are easy to use on the front-end and use a consumer grade interface which makes it easy for users to gather, deploy, integrate and consume information at a level scarcely imagined a decade ago.
The next step is then to automate these processes and move quickly from the collection of data to the presentation of insight and recommending actions. The true potential of both IIoT and AI technology comes from applying machine leaning and predictive analytics to a variety of practical and personalised use cases, whether it’s the farm that wants to project optimal field yield or the brewery that needs to estimate the amount of barley and hops to procure by month.
Pulling it all together
Driven by new data from the IIoT, AI has the potential to provide advice, discover performance patterns, analyse multiple influencing factors, and draw complex conclusions about a specific question — including questions that require a window into the future.
The maximum is reached when AI can emulate and enhance human performance, offering advice that is reliable and intelligent. This predictive insight helps organisations anticipate, understand and prepare for future trends and outcomes. But that insight is based on the IIoT sensors in the here and now.
A comprehensive IIoT/AI strategy should then automate the machine-learning process, continually refining the 20 or 30 different factors that might go into the algorithm in the face of new investment in better sensors or new sources of data.
This is especially important in food manufacturing as manufacturers race to develop new products based on changing customer preferences for non-GMO and additive-free foods. As ‘classic’ products fade into history and new buying habits are established, companies can turn to AI-driven analytics to monitor costs, optimise margins, and refine supply chain decisions. Elsewhere, line-of-business managers and plant operational teams can proactively address variables such as changing weather patterns, or sudden shifts in crews.
Rather than being overwhelmed by the massive amount of data presented by the IIoT, forward-thinking food manufacturers see it as the natural partner of AI-driven analytics that can be used to spot issues, opportunities or unknowns that simply would never be visible through traditional reports.
Mike Edgett is Industry & Solution Strategy Director, Process Manufacturing at Infor. Infor is a global leader in business cloud software specialised by industry. With 17,300 employees and over 68,000 customers in more than 170 countries, Infor software is designed for progress.